Research And Application Of Handwritten Chinese Character Segmentation

With the rapid development of computer information technology, the traditional paper texthas been unable to meet the information requirements, the paper text input through a variety ofscanning devices to the computer, is about the traditional paper text into digital information hasbecome an inevitable trend of times. With the continuous development of a variety of scanningequipment, OCR Optical Character Recognition system-based electronic devices has beengradually put into use .OCR technology is the process for the digital information of an image, including theoriginal image input, image preprocessing, character segmentation, character feature extractionand character recognition, a few steps. OCR’s ultimate goal is to paper information into digitalinformation, and character recognition, character segmentation in the process is a basis forcharacter recognition and difficult. In this paper, handwritten Chinese characters, the researchfocuses on the character segmentation and the refinement of character image in the imagepre-processing.Handwritten Chinese Character thinning and segmentation method is still a lot ofinadequacies, such as the character image thinning is not complete, misunderstanding of thecharacter image segmentation or over-segmentation phenomenon. Address these inadequacies,the original handwritten Chinese character thinning and segmentation method, a definiteimprovement, a better solution refinement is not complete, and misuse of segmentation.For binary images of handwritten Chinese characters, doing well to exclude the interferenceof other factors, the character image and background separation, pay more attention to the detailsof the character image, is conducive to the character image feature extraction and characterrecognition.The main work of this paper is as follows:(1) Describes the application and development of OCR technology, analysis and researchthe character segmentation research status for this article. Character feature for handwrittenChinese character and character features, analyzing the advantages and disadvantages of usingvarious methods, and comparing its segmentation accuracy.(2) Character image preprocessing, including image denoising, image binarization and textline segmentation. Including the working methods used and the benefits of the pretreatment, andthe corresponding experimental results compared to the need for pretreatment.(3) Discusses the characteristics of the characters of handwritten Chinese characters as wellas the uniqueness of the refinement of character images for handwritten Chinese characters,namely refinement of handwritten Chinese characters. Brief introduction to the characterrefinement of Research and research background for the refinement method to analyze an improved handwritten Chinese character refinement method based on mathematical morphology.(4) Describes two major difficulties in the word segmentation method, the adhesion wordrecognition and how to find the right split point. The inadequacies of the original connectedcomponent labeling and projection-based segmentation method, this paper presents an improvedprojection-based handwritten Chinese character segmentation method.(5) According to the method described in the paper, handwritten Chinese character imageexperiment, the experimental results were analyzed and compared. Elaborate on the direction ofthis study also shortcomings and areas for improvement, and describes the application ofhandwritten Chinese character segmentation in the field of pattern recognition.